LAW FIRMS GRAPPLING WITH COST OF LEVERAGING GENERATIVE AI IN E-DISCOVERY

"While a powerful technology, generative AI can also come with a hefty price tag. While some law firms are hopeful of lower bills down the line when using the technology in e-discovery, others aren't so sure yet".

Nearly a year after the deployment and availability of generative artificial intelligence, the legal industry has taken several approaches to experimenting with this new technology.

For e-discovery professionals, who face some of the most time-consuming and expensive tasks in the legal field, the technology has shown great promise. But one major obstacle stands in the way of its adoption: price.

A group of law firm professionals spoke at Relativity Fest 2023 in a session titled "Law Firms, Generative AI & The Future of E-Discovery" and discussed their experiences with the cost of generative AI, experimenting with various tools based on the technology.

The result was, "Cost makes the adoption of generative AI a hard sell."

Generative AI represents a breakthrough for the e-discovery process, mitigating the fallibility of human review and the shortcomings of technology-assisted review (TAR). However, this may not be a compelling enough argument for adoption of the technology.

Matt Jackson, data analytics and discovery consultant at Sidley Austin, said the cost of a tool is probably the most important indicator of adoption for the company, and generative AI has a high cost.

"We saw it early on with TAR, every new technology starts with a higher price tag and then, over time, the technology gets cheaper and better," Jackson said. "But I represent my customers.

My clients are cost-conscious. I can't go to them and say, "I want to use this really interesting tool, but it's going to cost you more, "it will never work."

In addition, e-discovery is often the most expensive part of legal matters, and although more and more clients and lawyers are coming around to the idea that generative AI integrations can reduce overall bills over time, the upfront cost is what makes it "hard to sell," Jackson noted.

Foley & Lardner: "Generative AI will reduce e-discovery costs over time."

In the end, decisions about adopting any technology must consider a cost-benefit analysis.

Michael Cichy, managing director of litigation support at Foley & Lardner, said this forward-looking approach is what his firm hopes to adopt when it comes to generative AI tools for e-discovery.

In addition, when a generative AI-powered solution makes a particular e-discovery decision, it offers the rationale "why it believes the document is responsive," whereas in "conventional review you don't get [this] information."

This technology can also detect a missed error early in the e-discovery process, reducing the downstream costs of changing an error after the fact.

"So even though the initial impact might be large, over time you could still fix the problem and you could get downstream savings," he added.

Troutman Pepper Hamilton Sanders: "The TAR tax is now the tax on generative AI."

It is difficult to analyze how e-discovery costs will develop with generative AI implementations, without taking into account court decisions on the use of this technology in legal work.

For example, Jason Lichter, director of Troutman Pepper eMerge, a subsidiary of Troutman Pepper Hamilton Sanders, noted that a substantial cost of generative AI is that imposed by the other side.

In the past, when opposing counsel challenged the use of a technology, in many ways it was a tactic that forced lawyers to spend more money and time to prove their case in favor of using the tool.

There was talk of a "TAR tax." But there should not be this additional obligation to be more transparent than there would be with linear review," Lichter said. "It may be necessary for a judge to first issue an opinion certifying that generative AI is presumptively reasonable or defensible, subject to appropriate conditions."

Ultimately, it may take time for the true cost of generative AI in e-discovery to come to light.

Melissa Dalziel, legal counsel at Quinn Emanuel Urquhart & Sullivan, pointed out that while she has found the obvious advantages of generative AI tools over TAR, continuous active learning (CAL or TAR 2.0) and even human review, and that her firm intends to continue experimenting with the technology, the costs and its impact on the firm's clients are yet to be seen.

The difficult part is, in the short term, "Will we save money by using AI?

Will the cost be the same?" We don't have the metrics to be ab§°*we can participate in the cost exercise so we can gather our metrics on 'Is it useful? Is it taking less time? Is it getting better results?"

But until then, it will have to be an iterative process in which all the generative AI pioneers in e-discovery will share their results to eventually develop quantitative cost metrics.